Aps AECDM MCP Dotnet – AI‑Enabled Access to Autodesk’s AEC Data Model

The Aps AECDM MCP Dotnet server bridges the gap between Claude Desktop and Autodesk’s AEC Data Model (ADM) API. By exposing a small set of carefully chosen tools through the Model Context Protocol (MCP), it allows developers to write natural‑language queries that automatically translate into authenticated API calls against an organization’s AEC hubs, projects, and design elements. The server handles OAuth 2 PKCE token acquisition, data retrieval, and even viewer rendering—all without requiring the client to manage credentials or API endpoints.

What Problem Does It Solve?

In traditional workflows, pulling information from the ADM requires a custom integration: developers must authenticate with APS, construct REST requests, parse JSON responses, and then feed the data into a UI or analysis pipeline. This process is repetitive and error‑prone, especially when multiple applications need to share the same data. The MCP server removes this boilerplate by presenting a unified, AI‑friendly interface. Claude can ask for “the list of element groups in project X” or “highlight all structural beams in the viewer,” and the server will automatically fetch, filter, and return the requested data or visual cues.

Core Capabilities

• Token Management – The tool handles PKCE flow, returning a short‑lived bearer token that the server uses for all subsequent APS requests.
• Hierarchical Data Retrieval – , , and expose the ADM hierarchy, enabling conversational exploration of an organization’s structure.
• Filtered Element Queries – lets users narrow results by category (e.g., “windows” or “columns”), supporting precise data extraction.
• Viewer Interaction – loads a design into the Autodesk Viewer, while can emphasize specific elements on demand.
• MCP‑Ready – Each tool is registered with the MCP SDK, ensuring compatibility with any client that implements the protocol (Claude Desktop, ChatGPT, or custom front‑ends).

Real‑World Use Cases

• Design Review Automation – An architect can ask Claude to “show me all openings in the first floor” and instantly see highlighted elements in the viewer.
• BIM‑Based Analytics – Data scientists can pull element counts by category and feed them into a predictive model, all through conversational queries.
• Project Status Dashboards – Project managers can query “which projects are overdue for delivery” and receive a concise list without touching the APS console.
• Education & Training – Students learning BIM can experiment with natural‑language commands to explore model data, reinforcing concepts in a hands‑on way.

Integration into AI Workflows

Because the server follows the MCP specification, any client that understands the protocol can tap into these tools. In practice, a developer would:

1. Register the server in Claude Desktop’s configuration (or another MCP‑compliant UI).
2. Invoke a tool via a natural‑language prompt (e.g., “GetProjects”).
3. Receive a structured response that can be further processed or visualized.

The server’s design eliminates the need for custom adapters, streamlining the development cycle and reducing maintenance overhead.

Unique Advantages

• Single‑Source Authentication – PKCE tokens are refreshed automatically, so developers never expose secrets.
• Domain‑Specific Tooling – Unlike generic HTTP clients, the server’s tools are tailored to ADM concepts (hubs, projects, element groups), making prompts more intuitive.
• Viewer Integration – Real‑time rendering and highlighting directly from the AI prompt give a powerful visual feedback loop.
• Cross‑Platform Support – Built on .NET 8, the server runs on Windows or macOS, matching the environments most developers already use.

In summary, the Aps AECDM MCP Dotnet server turns complex BIM data access into a conversational experience, enabling developers to focus on higher‑level logic while the server handles authentication, data retrieval, and visualization.